Functional Outcome Prediction in Young Adults With Mental Health Symptoms Using Machine Learning and Large Language Models: Longitudinal Observational Study.
Authors
Affiliations (25)
Affiliations (25)
- Department of Psychiatry and Psychotherapy, Carl Gustav Carus University Hospital, TUD Dresden University of Technology, Dresden, Germany.
- Department of Psychiatry and Psychotherapy, Jena University Hospital, Philosophenweg 3, Jena, Thuringia, 07743, Germany, 49 36419300.
- Department of Psychiatry, University of Marburg, Marburg, Germany.
- Core-Facility Brain Imaging, Faculty of Medicine, University of Marburg, Marburg, Germany.
- Center for Mind, Brain and Behavior (CMBB), University of Marburg and Justus Liebig University Giessen, Marburg and Giessen, Germany.
- Department of Psychiatry and Psychotherapy, University Medical Center Hamburg-Eppendorf, Hamburg, Germany.
- Centre for Psychiatry, Justus-Liebig University Giessen, Giessen, Germany.
- Tübingen Center for Mental Health, Department of Psychiatry, University of Tübingen, Tübingen, Baden-Wurttemberg, Germany.
- Partner Site Tübingen, Deutsches Zentrum für Psychische Gesundheit, Tübingen, Germany.
- Biomedical Magnetic Resonance, University of Tübingen, Tuebingen, Baden-Wurttemberg, Germany.
- Vivantes Hospital Am Urban and Vivantes Hospital Im Friedrichshain, Department of Psychiatry, Psychotherapy and Psychosomatic Medicine, Charité-Universitätsmedizin Berlin, Berlin, Germany.
- Department of Psychiatry, Psychosomatic Medicine and Psychotherapy, University Hospital, Goethe University Frankfurt, Frankfurt, Germany.
- Cooperative Brain Imaging Center - CoBIC, Goethe University Frankfurt, Frankfurt, Hesse, Germany.
- Institute for Translational Psychiatry, University of Münster, Münster, North Rhine-Westphalia, Germany.
- Charité Campus Mitte, Department of Psychiatry and Psychotherapy, Charité University Medicine, Berlin, Germany.
- Department of Clinical Psychiatry and Psychotherapy, Brandenburg Medical School Theodor Fontane, Neuruppin, Brandenburg, Germany.
- Department of Child and Adolescent Psychiatry, Charité Universitätsmedizin Berlin, Berlin, Germany.
- Department of Psychiatry, Northwell Health, Zucker Hillside Hospital, New York, NY, United States.
- Department of Psychiatry and Molecular Medicine, Donald and Barbara Zucker School of Medicine at Hofstra/Northwell, Hempstead, NY, United States.
- Department of Psychiatry, Psychotherapy and Preventive Medicine, LWL University Hospital, Ruhr-University, Bochum, Germany.
- Else Kroener Fresenius Center for Digital Health, TUD Dresden University of Technology, Dresden, Germany.
- Institute for Translational Psychiatry, University of Münster, Münster, Germany.
- Department of Psychiatry, Medical School and University Medical Center OWL, Protestant Hospital of the Bethel Foundation, Bielefeld University, Bielefeld, Germany.
- German Center for Mental Health (DZPG), Jena Magdeburg Halle, Germany.
- Center for Intervention and Research on Adaptive and Maladaptive Brain Circuits Underlying Mental Health (C-I-R-C), Jena Magdeburg Halle, Germany.
Abstract
Functional impairments associated with mental health conditions are on the rise. Predicting functional outcomes may improve the targeting of preventive interventions. While prognostic models have primarily focused on psychosis, early recognition services require a transdiagnostic approach. This study aimed to predict global functioning within a 2-year follow-up using baseline clinical and structural magnetic resonance imaging (MRI) data in a population-based sample of young, help-seeking individuals presenting with affective and anxiety symptoms as well as attention-deficit hyperactivity disorder. We classified 357 help-seeking individuals aged 18-35 years recruited from 9 sites as "impaired" (Global Assessment of Functioning [GAF] ≤60; n=228) or "nonimpaired" (GAF>60; n=129) at year 1 and/or year 2 follow-up. GAF classification group status at follow-up was predicted using linear support vector machine (SVM), decision tree, and large language model (LLM) Llama-3 using clinical assessments and/or structural MRI. Leave-one-site-out (SVM) or external sample (LLM) was used for validation. SVM achieved balanced accuracy of 69.2% using clinical features only. Items related to baseline occupational functioning, interpersonal relationships, cognitive functioning, psychotic and affective symptoms, as well as the presence of anxiety disorder, were most predictive. The decision tree further reduced the feature set to 5 predictive items, achieving balanced accuracy of 76.6%. Although amygdala and hippocampal subregions achieved balanced accuracy of 57.1%, structural MRI did not improve the overall prediction. Llama-3 performed comparably well to SVM (balanced accuracy of 72.6%). Machine learning demonstrated good performance in predicting global functioning. Interestingly, the out-of-the-box LLM performed comparably well without being trained or fine-tuned, highlighting the potential of leveraging free-text data for mental health prognosis.